i am using the Sentinel 3 SYN data to retrieve biophysical variables. I would like to understand how cloud flagging works, I encountered some issues.
I came across a site next to Antwerp, where the cloud CLOUD_flags_CLOUD_MARGIN tends to classifiy water as clouds (at least this is what I see).
see the attached images.
now I understand that it might be a cloudy day. However, I am currently processing more than 3 years of data over the same site, and unfortunately this “cloud” is there most clear days as well, indicating some sort of error.
Can you please help me out to understand this issue, and whether or not I should be using the cloud mask?
the bounding box for the site is: [ 4.4776028075, 51.2811759829, 4.560966531, 51.3318354764 ]
The cloud flag, you are using, is the same cloud flag, that is used in OLCI L2 processing.
I am the main developer of the OLCI cloud detection algorithm and hopefully can help you. The current OLCI cloud flagging algorithm is based on a neural network and is validated with an overall accuarcy of 86%-88%. We are aware of some sytematic issues, but those are normally related to coastlines and bright targets. Therfore, this seems a bit odd.
I checked the site with google and found a fixed installation in the water (see image below), which might cause a constant false cloud flagging. Since OLCI is not on a fixed grid like Sentinel-2 MSI, this could move around one pixel over time.
I’m following the discussion and have a question. If this false cloud flagging keeps happening because of features in the water, is there a way to improve the cloud mask to handle it better? Or should we check manually for places like this?
